Graph Construction Method for GNN-Based Multivariate Time-Series Forecasting

Multivariate time-series forecasting(MTSF)plays an important role in diverse real-world applications.To achieve better accuracy in MTSF,time-series patterns in each variable and interrelationship patterns between variables should be considered together.Recently,graph neural networks(GNNs)has gained much attention as they can learn both patterns using a graph.For accurate forecasting through GNN,a well-defined graph is required.However,existing GNNs have limitations in reflecting the spectral similarity and time delay between nodes,and consider all nodes with the same weight when constructing graph.In this paper,we propose a novel graph construction method that solves aforementioned limitations.We first calculate the Fourier transform-based spectral similarity and then update this similarity to reflect the time delay.Then,we weight each node according to the number of edge connections to get the final graph and utilize it to train the GNN model.Through experiments on various datasets,we demonstrated that the proposed method enhanced the performance of GNN-based MTSF models,and the proposed forecasting model achieve of up to 18.1%predictive performance improvement over the state-of-the-art model.

A systematic correlation between morphology of porous carbon cathode and electrolyte in lithium-sulfur battery

Porous carbon has been applied for lithium-sulfur battery cathodes,and carbonized metal-organic framework(MOF)is advantageous in tuning the morphology.Herein,we have systematically synthesized water-distorted MOF(WDM)derived porous carbon via controlling the proportion of both water in a mixed solvent(dimethylformamide and water)and ligand in MOF-5 precursors(metal and ligand),which is categorized by its morphology(i.e.Cracked stone(closed),Tassel(open)and Intermediate(semi-open)).For example,decrease in water and increase in ligand content induce Cracked stone WDMs which showed the highest specific surface area(2742-2990 m^(2)/g)and pore volume(2.81-3.28 cm^(3)/g)after carbonization.Morphological effect of carbonized WDMs(CWDMs)on battery performance was examined by introducing electrolytes with different sulfur reduction mechanisms(i.e.DOL/DME and ACN_(2) LiTFSITTE):Closed framework effectively confines polysulfide,whereas open framework enhances electrolyte accessibility.The initial capacities of the batteries were in the following order:Cracked stone>Intermediate>Tassel for DOL/DME and Intermediate>Tassel>Cracked stone for ACN_(2) LiTFSI-TTE.To note,Intermediate CWDM exhibited the highest initial capacity and retained capacity after 100 cycles(1398 and 747 mAh/g)in ACN_(2) LiTFSI-TTE electrolyte having advantages from both open and closed frameworks.In sum,we could correlate cathode morphology(openness and pore structure)and electrolyte type(i.e.polysulfide solubility)with lithium-sulfur battery performance.